This article was downloaded by: [] On: 03 September 2014, At: 00:42 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tres20 15 years of processing and dissemination of SPOT-VEGETATION products Bart Derondea, Walter Debruyna, Eric Gontiera, Erwin Goora, Tim Jacobsa, Sara Verbeirena & Johan Vereeckena a VITO Remote Sensing Unit, Mol, Belgium Published online: 27 Mar 2014.

To cite this article: Bart Deronde, Walter Debruyn, Eric Gontier, Erwin Goor, Tim Jacobs, Sara Verbeiren & Johan Vereecken (2014) 15 years of processing and dissemination of SPOT- VEGETATION products, International Journal of Remote Sensing, 35:7, 2402-2420, DOI: 10.1080/01431161.2014.883102 To link to this article: http://dx.doi.org/10.1080/01431161.2014.883102

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions Downloaded by [European Space Agency] at 00:42 03 September 2014 International Journal of Remote Sensing, 2014 Vol. 35, No. 7, 2402–2420, http://dx.doi.org/10.1080/01431161.2014.883102

15 years of processing and dissemination of SPOT-VEGETATION products Bart Deronde*, Walter Debruyn, Eric Gontier, Erwin Goor, Tim Jacobs, Sara Verbeiren, and Johan Vereecken

VITO Remote Sensing Unit, Mol, Belgium (Received 29 October 2012; accepted 25 March 2013)

Throughout the VEGETATION programme, the Flemish Institute for Technological Research (VITO) uninterruptedly hosted the prime user segment of both VEGETATION 1 and VEGETATION 2 multispectral instruments on board the Satellite Pour l’Observation de la Terre 4 (SPOT 4) and SPOT 5 satellites. Operational since the launch of SPOT 4 in March 1998, and foreseen to continue at least until the end of the SPOT 5 mission (anticipated in spring 2014), this user segment comprises a processing facility (PF), actively receiving, processing, correcting, archiving, and dis- tributing the VEGETATION data and derived added-value products. First and foremost, the VEGETATION programme has been serving the needs of operational users – both institutional and commercial – requesting data in near-real time. However, scientific and educational users too benefited significantly, in particular from VEGETATION’sunique time series of the Earth’s land cover, and more specifically the vegetation cover. Over the years, the centralized archive houses processed data covering the equivalent of 11,000 times the Earth’s surface, and delivered more than 50 terapixels to around 10.000 users. As such, VEGETATION’s mission is a prime example of what Europe wants to achieve through the Global Monitoring for Environment and Security (GMES) initiative: truly operational services providing reliable and up-to-date information. This article describes the processing facility, the way the data and products are archived, the different dissemination channels as well as the data policy adopted and the users served. One of the recent evolutions, the development of an entirely new product distribution facility (PDF), implemented as part of the Project for On-Board Autonomy – Vegetation (PROBA-V) user segment is discussed.

1. Introduction Downloaded by [European Space Agency] at 00:42 03 September 2014 We only have one planet –‘Spaceship Earth’. It is our sole heritage … all we have to pass on to our children. And it deserves to be treated as such. (http://spot4.cnes.fr/spot4_gb/spot4.htm)

In response to this challenge, the VEGETATION programme has been conceived to monitor environmental change – and especially the evolution of the world’s vegetation cover – on a global and daily basis. The programme is the fruit of a collaboration between various European partners: Belgium, France, Italy, Sweden, and the European Commission. In 1998, it was grafted onto the Satellite Pour l’ObservationdelaTerre (SPOT) programme, founded by Belgium, France, and Sweden in 1978. The VEGETATION programme consists of two observation instruments, VEGETATION 1 (VGT 1) and VEGETATION 2 (VGT 2), as well as all necessary ground infrastructures.

*Corresponding author. Email: [email protected]

© 2014 Taylor & Francis International Journal of Remote Sensing 2403

The first of the two instruments in orbit is aboard the SPOT 4 satellite, launched on 24 March1998(Henry1999). The second is aboard SPOT 5, launched on 4 May 2002. On 29 April 2012, VGT 1 was switched off after more than 14 years of uninterrupted observations. Its platform, SPOT 4, was drifting to a lower orbit and earlier overpass time, while its prime instruments, the two High-Resolution Visible and Infrared (HRVIR) sensors remained active until December 2012. VGT 2 is still active and is planned to continue its observations until 31 May 2014 (Swinnen et al. 2014).

2. The processing facility Since the launch of SPOT 4 and the subsequent launch of SPOT 5, the processing facility at the Flemish Institute for Technological Research (VITO) (also called CTIV – Centre de Traitement d’Images VEGETATION) in Mol, Belgium, has been processing the daily VEGETATION data stream. The data is sent via X-band from the satellites to the receiving station in Kiruna, Sweden. This station is located at a high latitude, which enables a large number of daily overpasses (owing to the quasi-polar orbit) and thus offers multiple possibilities to download the mass memory. Indeed, the memory capacity corresponds to about 90 minutes of observations and needs 4–5 downloads every day. The data transmission on the ground is carried out through an X-band link (8.2 GHz). A first preprocessing of the data is realized in the receiving station; it consists of the proper recovery of the viewing segments, and in the provision of auxiliary data (orbit, date, satellite attitude data, …), which are to be used during the processing sequence. The CTIV is operated by VITO. It receives the data pre-processed by the VEGETATION X-band imagery receiving station in Kiruna through a dedicated high- speed link, which is activated after each data reception by the station. The role of the CTIV is, on the one hand, to ingest, process, and archive all VEGETATION data (systematic tasks) and, on the other hand, to provide a catalogue, production, and delivery service to the user community. The CTIV is operated seven days a week during regular office hours. A high degree of automation allows the centre to be operational continuously. This includes acceptance, processing, and delivery of user orders (Fierens and Van Speybroeck 2001). The centre is designed to deliver products less than 48 hours after image acquisition, on condition that the user has already ‘sub- scribed’ to imagery covering the area of interest. All processing is carried out as soon as the necessary raw image telemetry data and auxiliary data, such as meteo, are available, with as little human intervention as possible. Downloaded by [European Space Agency] at 00:42 03 September 2014 2.1. The VEGETATION systematic processing chain Figure 1 provides an overview of the main processing algorithms used in the VEGETATION image processing facility. Starting from the telemetry data received from the receiving station to the final products delivered to the users, the geometric processing, the radiometric corrections, the atmospheric corrections, the synthesis composition, and the product formatting are the main steps of this process. The automated processing chain transforms the incoming VEGETATION raw data into products with added value. At the end of the processing chain, the following standard VEGETATION products are available to the user community.

● P or Physical products: zone extracted from a single orbit, geometrically and radiometrically corrected. 2404 B. Deronde et al.

GCP (SPOT4) Poles coordinates (SPOT5) Reception of Standard Map VEGETATION Geomodelling Atomospheric MVC Synthesis Product creation (Plate correction Generation Data via X-band Radiomodelling BDC Synthesis Carée, 1 km) (SMAC ) (P, MVC, BDC) Preprocessing Kiruna PCI (QIV) METEO

Archive SUV catalogue

CTIV

Figure 1. Overview of the SPOT-VEGETATION image processing chain.

● S or Synthesis products: ‒ S1 or daily synthesis: synthesis of all data acquired over a 24 hour period. ‒ S10 or 10 day synthesis: a result of the merging of data strips from 10 consecutive days. ‒ D10: 10 day synthesis based on a bi-directional reflectance distribution function. ‒ S-NDVI products: contain only the NDVI (normalized difference vegetation index). ‒ S-total products: contain the NDVI, all spectral bands, and acquisition parameters.

2.2. Radiometric calibration The operational absolute radiometric calibration was originally based on gains and offsets as derived from irradiance measurements of an on-board calibration lamp (Meygret 1998, 2000; Henry and Meygret 2001). Later, owing to degradation of the on-board calibration lamp, the operational radiometric calibration shifted to vicarious in-orbit techniques based on systematic (monthly) acquisitions over radiometrically homogeneous stable desert sites located in the Sahara. The vicarious calibration result thus obtained is regularly validated via the Rayleigh calibration approach and inter-band calibration acquisitions over deep convective clouds and Sun-glinted areas. Dark current calibration is performed by image acquisitions over during the night. This operational calibration procedure is Downloaded by [European Space Agency] at 00:42 03 September 2014 performed by the QIV (the Image Quality Centre at the Centre National d’Etudes Spatiales (CNES) in Toulouse, France), which subsequently provides the CTIV with updated calibra- tion coefficients to be implemented in the VEGETATION processing chains.

2.3. Radiometric recalibration As explained above, the radiometric calibration depended primarily on on-board calibra- tion devices. In 2006, a cross-calibration campaign (VGT 2 vs. POLDER (Polarization and Directionality of the Earth’s Reflectances) 2, POLDER 3, and VGT 1) over deserts indicated that the VGT 2 calibration lamp was not evolving as expected. The inter- comparison showed that the on-board calibration system had systematically overestimated the sensitivity losses in some spectral bands. The VGT 2 calibration coefficients in these bands were therefore lower and the provided reflectance values were higher than they International Journal of Remote Sensing 2405

should have been. Therefore, the complete VGT 2 archive was reprocessed with corrected calibration coefficients based on the vicarious approach. This campaign was completed at the beginning of 2007 (Bartholomé et al. 2006). During 2009 and 2010, the VGT 1 archive was also reprocessed based on the vicarious approach to maintain consistency over time for all the VGT 1/VGT 2 reflectance data. All P, S1, and S10 products have been recalculated besides part of this campaign. In March 2012, a harmonization study on processing methods and products revealed an anomaly in the VEGETATION processing chain. This artefact concerns the incorrect implementation of the standardization of solar illumination. This standardization has a direct impact on the values of the top-of-atmosphere reflectance provided in the VGT-P products, as well as secondary effects on the atmospherically corrected S1 and S10 products. The correction of this anomaly will be performed during a reprocessing cam- paign after the end of the VEGETATION mission, which will last until May 2014. One key objective of this reprocessing operation is ensuring full consistency with the proces- sing chain for Project for On-Board Autonomy – Vegetation (PROBA-V), the VGT successor scheduled for launch in early 2013. The calibration approach for PROBA-V is described in, e.g., Sterckx et al. (2014); a consistent inter-calibration campaign is to be established during the commissioning phase of PROBA-V.

2.4. Geometric calibration The VEGETATION image geometry is corrected differently for VGT 1 and VGT 2. For the VGT 1 instrument, the correction is implemented in two steps. In a first step, the gyroscopic measurements, which are part of the SPOT 4 telemetry, are converted into corrections for changes in the satellite’s attitude (corrections for roll, pitch, and yaw). In a second step, an absolute reference attitude is assigned to the start of each orbit. This is accomplished by comparing discernible reference points in the imagery with a database of ground control points. The correlation of image points and ground control points is used to derive the absolute attitude and position of the satellite at the start of every orbit. The assignment of the absolute satellite attitude, as described, is a semi-automatic process. Operator intervention is still required for selecting an adequate distribution of cloud-free reference points in the imagery. The subsequent correlation and derivation of the corrective parameters are automated pro- cesses (Fierens and Van Speybroeck 2001; Sylvander et al. 2000; Passot 2001). For the VGT 2 instrument on board SPOT 5, a star tracker provides a reliable estimation of the satellite attitude, implying an accurate absolute image location: it is therefore not necessary to further improve VGT 2 imagery using ground control points Downloaded by [European Space Agency] at 00:42 03 September 2014 (Sylvander, Albert-Grousset, and Henry 2003).

2.5. Map projection in Plate-Carrée The first step of the synthesis processing is transforming all segments to the same cartographic projection. The chosen projection is Plate-Carrée with 1/112° per pixel in line and row, which is about 1 × 1 km at the equator, and 0.5 km wide and 1 km long at 60° latitude. Hence, previously computed geometrical grids are used, related to a digital elevation model (Earth Topography 5-minute (ETOPO5)) to compute the location of the pixels taking into account the relief effects. This is particularly important on the edges of the viewing field; an interpolation is performed between the two location grids at 0 m and 5000 m elevation. To go from the ‘system projection’ coordinates to the Plate Carrée coordinates, the cartographic coordinates are computed using the Geolib Institut 2406 B. Deronde et al.

Géographique National (IGN) library. To calculate the best radiometry of the resulting pixel, a bi-cubic interpolation is performed using a 4 × 4 window, from the 16 nearest pixels in the raw image. The products resulting from these steps are the so-called VGT-P products. These products are adapted for scientific applications requiring highly accurate physical measurements and contain top-of-atmosphere reflectance (TOA).

2.6. Derivation of ground reflectance The main objective of the VEGETATION processing chain is providing products from which bio-geophysical parameters can be derived. Therefore, atmospheric influences should be eliminated from the imagery, thus yielding ground reflectance values. The atmospheric corrections applied make use of near-real-time global meteorological data. This global meteorological dataset is generated four times daily. The corrective software used is the Sequential Model-based Algorithm Configuration (SMAC) developed by Rahman and Dedieu (1994), which in turn is based on the 6S model. New algorithms for eliminating the effects of tropospheric aerosols and bi-directional effects (BDC) were introduced in 2000 (Duchemin et al. 2000; Maisongrande et al. 2001; Duchemin et al. 2001; Duchemin and Maisongrande 2002). Consequently, all segments captured in one day are combined into a world synthesis (S1 product), according to the maximum value compositing principle, i.e. the pixels selected for the synthesis are based on the selection of the maximum NDVI value in all available segments, to ensure coverage of all landmasses worldwide with a minimal effect of cloud cover.

2.7. Cloud screening and filtering Cloud screening and filtering is performed in two steps. In a first step, a cloud detection algorithm based on the spectral signature of the individual pixels flags them as cloudy, not cloudy, or not sure (Lissens et al. 2000). In a second step, the global S1 datasets of 10 consecutive days are compared pixel by pixel and, based on cloud flags and maximum NDVI value, the least cloudy pixel of these 10 consecutive days is selected and included in a 10-day synthesis dataset (S10). This dataset, consisting of a global 1 km cloud-free atmospherically corrected top-of-canopy reflectance map in the standard VEGETATION projection, is the basis for all VGT-S10-derived user products. All S1 and S10 products are subject to visual inspection by CTIV operators (e.g.

Downloaded by [European Space Agency] at 00:42 03 September 2014 Fierens and Van Speybroeck 2001). The information content of a standard S10 product is provided in Table 1.

3. The VEGETATION archive 3.1. Archived datasets All data transmitted by the VEGETATION instruments, i.e. the raw image telemetry, are archived on ‘Linear Tape-Open’ LTO-3 tapes. The correction parameters as calculated by the CTIV processing chains are stored together with the image data. The corrective parameters include both radiometric and geometric correction factors of the image data. These data are backed up twice on an LTO-3 tape (one archive tape and one clone tape, which is stored at a separate remote location (safeguard)). The corresponding meteorolo- gical data are also archived (e.g. Fierens and Van Speybroeck 2001). International Journal of Remote Sensing 2407

Table 1. Content of a standard VEGETATION S10 product.

Acronym Parameter Description

PHYS_VOL Physical volume descriptor Leader file of the product, containing some basic information and the list of files available in the product LOG Logical volume descriptor Information about • map projection information (general information, geodetic system parameters, projection parameters) • cartographic location • geographic location • image coordinates (corresponding to carto- and geographic location) • geometric correction • radiometric correction • orbit parameters • date and time • algorithms references RIG Copyright descriptor Information about the copyright BO B0 spectral band Radiometry data : 0.43–0.47 µm B2 B2 spectral band Radiometry data : 0.61–0.68 µm B3 B3 spectral band Radiometry data : 0.78–0.89 µm MIR MIR spectral band Radiometry data : 1.58–1.75 µm NDV Vegetation index Normalized difference vegetation index (NDVI) SM Status map Info on radiometric quality of spectral bands, presence of ice/snow, cloud, etc. VZA Viewing zenith angle Data on the satellite zenith angles VAA Viewing azi